[GW2] Back! And Marionette Success Rates

Hello Tyria!

We're back with a blast. After over a year without MMOs, I decided to take a quick check at what was new in GW2.

... and quickly wandered into another data-driven gaming argument.

In the Origins of Madness update, ArenaNet introduced two world raid bosses: The Twisted Marionette and a three headed Great Jungle Wurm. While the encounters have difficult mechanics, they're part of the open world - you simply cannot choose the other players you'll fight alongside. Since each boss has mechanics that make it possible for a single player to cause the whole server to fail the encounter, players have taken to complaining about the mechanics of both fights on the forums (shocker!).

I'm particularly interested in the Marionette because of this comment, made by Josh Foreman (Environment Design for ANet), in responding to complaints about the fight's difficulty:

if after a week the Marionette is still only being beaten 1 out of 10 times, then I would say we may have tuned it a bit too difficult.

That was 14 days ago. A week after he made that comment, I started tracking the success of the Marionette event via the GW2 API. The Events section of the api tracks the status of each event on every server, noting its outcome (success/fail). This gives me six days of data on Marionette kill attempts. Since the Marionette fight occurs once every two hours, that is data on just over 70 Marionette attempts.

The overall Marionette success rate for the past week is 6.6%.

• on North American servers the success rate is 6.1%.
• on European servers the success rate is 7.0%.

Server Population
One issue with the Marionette is that the fight takes a large number of players to be successful. The fight is divided into five lanes, which are periodically divided further, into five platforms. Most guides for the fight mention 20-25 players per lane (100-125 players for a successful attempt), though a coordinated group could get by with less. So active, well-populated, coordinated servers should do better on this boss.

There are vast differences between servers.  The top EU server (Desolation) succeeds on 63% of its attempts, and the top NA server isn't far behind (Blackgate - 59%). On the other hand, I have never observed a successful Marionette attempt on half of all servers (25 of 51 servers).

As might be expected, servers with a higher WvW rank are much more successful at the Marionette. While top 5 WvW servers average a greater than 20% success rate, all other servers have a less than 5% sucess rate at the event.

Prime Time
In addition to vast differences between servers, it matters what time the event is being attempted. NA primetime (starting at 7 EST) has a much higher success rate than afternoons (students getting off school?).

With this in mind, I have two caveats:

1. GuildWars 2 has an overflow system, where if too many players are in a zone, an overflow server is created for any additional players that arrive (or several overflows, if need be). I don't have any information about what happens on these servers (or how many there are).
2. It could be that the GW2:Events API is not accurate. Completely unsuccessful servers are pretty shocking, so much so that I doubt the source. Then again, I do the Marionette at least once per day, and the outcome of each of those attempts is accurate, when I check against the data. 

Further Questions
So what does this mean for events like the Marionette?

Feedback based on a particular server (or time) isn't really representative of the experience of the event as a whole.  Someone trying the event at 7 or 9 pm EST on Blackgate is virtually guaranteed a successful run. But there are 25 servers that I've never seen succeed at the event.

On the other hand, vast differences between servers (and times) suggests that the fight isn't balanced across the wide range of situations GW2 players are likely to encounter. Hopefully, future events will be open to a wider range of the community, instead of being limited / exclusive content.

Time willing, I'll take another look at the Marionette sometime this week, and see if I can break this apart even more.